EMC simulation addresses ECU validation issues

Transcription

1 EMC simulation addresses ECU validation issues A more straightforward validation of electromagnetic compatibility can be achieved by combining tools. By Stefan Heimburger, Andreas Barchanski, and Thorsten Gerke The growing electrification of automotive platforms would have made Thomas Edison happy, but for automotive product development groups it means all sorts of new EMC validation challenges. Driven by the growing number of electrical functions such as drive systems for hybrid vehicles or pure electrical drives, driver assistance systems, and advancements in consumer electronics, electrification of automotive platforms means teams have to validate each systems for electromagnetic compatibility (EMC) across the complete operating range to ensure full compliance with the required quality and safety criteria. In particular the noise emission of dedicated components or embedded control units (ECU) may adversely impact the vehicle's network or other components. EMC specifications determined by EU framework directive 2007/46/EC or according to rule ECE-R10 require careful verification. 1, 2 Verification tests span frequency ranges up to 2GHz or higher. To complicate matters, vehicle manufacturers often impose additional EMC constraints on the system designs. Identifying and resolving design malfunctions caused by EMC issues can consume significant resources and add cost intensive design iterations. Often, ECU implementation-specific problems are encountered that require answers to the following questions. Whether a lead frame has to be used or an ordinary circuit board? Which HF filter types are being deployed? What about the control unit construction and the positioning of the components within the ECU? Should the emission be analysed as narrowband or broadband? In extreme cases, it may be necessary to change the physical design, the ECU architecture, or filter elements. Implementing these changes may be costly and require additional development time adding risk to platforms that are otherwise ready for production. Substantial changes may even delay the product launch. In order to minimise the risk associated with these issues, early detection is crucial. To achieve this goal, designers are employing simulation tools to develop virtual solutions. Simulation-based methodologies enable engineers to perform EMC analyses of the control unit early in the development process, even before the ECU is available for EMC testing. 3 Typically, applying the EMC tests to automotive ECUs in actual hardware is time consuming. In addition, it's difficult to precisely reproduce the exact measurement conditions such as temperature or device parameter drift. Owing to the persistent miniaturisation of automotive electronic components, direct measurements might not even be possible to perform. In such scenarios, simulation presents the only path to validate EMC performance. Combining 3D field simulation and system simulation with discrete components Creating a simulation model is critical for the success of simulation-based approaches. Understanding the advantages and limitations of the available modelling and simulation approach plays an important role. Partitioning of the simulation model provides a possibility to study the EMC behaviour of a complex automotive ECU system. First, the ECU system is decomposed into smaller components such as the circuit board, plug connectors, and cables. By means of 3D field simulation, these parts can be analysed separately to determine their individual electromagnetic performance. Afterwards, the EM field simulation data can be used to extract discrete models of the components, which are connected in a system simulator at the second stage. In a 3D field simulation, Maxwell's equations are solved within a given volume by applying numerical methods. The user provides the geometry to be analysed, such as the lead frame of the ECU power electronics, and defines stimulus ports. Typical outputs from the field simulator include scattering parameters (s-parameters) and geometric distributions of currents and fields. In particular the ability to visualise the EM fields or the current distribution can be helpful because it provides insight that typically cannot be obtained by a measurement. Moreover, the field data provides information that can be used to identify potential coupling paths within the ECU physical structure. Typical applications include calculation of return current distribution on the reference layer of a circuit board, coupling of fields from mounted inductors, resonance generation in enclosures, or transmission behaviour of connectors. If the system components are spatially separated from each other and don't interact through radiated emission-- like a sensor wired to an ECU--it doesn't pay to model and simulate the complete system within a 3D field solver. In this case, the computational effort and required simulation time would increase significantly and swamp the benefits associated with minimal accuracy improvements. Combining the complementary advantages of system simulation and 3D field simulation provides a good approach for this problem. Precise EMC models of the individual control device components can be generated using a 3D field simulation within the CST Studio Suite and ported to discrete models that are used by a system simulator like Saber. This approach produces a holistic system model for EE Times-Asia eetasia.com Copyright 2013 emedia Asia Ltd. Page 1 of 5

2 EMC simulation of the ECU without requiring a 3D simulation of the complete ECU system. The design partitioning approach combines the strengths of both simulation models, preserving the accuracy of the EMC component models and the efficiency and coverage of a system model. In addition, working at the system level enables the use of other behavioural models to simulate the effects of logic, regulation, and control algorithms required for pulse-width modulation (PWM). These sub-systems can be modelled using VHDL-AMS or other description languages. In this way, the electromagnetic emission behaviour of the ECU system can be analysed precisely. The need for behavioural models often reaches beyond control sub-systems. Saber includes a comprehensive library of models such as power transistors or insulated-gate bipolar transistors (IGBTs) that designers can use. These models can be combined and simulated together with the EMC models generated by the 3D field simulation. In order to provide designers with a seamless link between Saber and the CST Studio Suite, Synopsys and CST have developed an interface which allows an almost seamless model exchange from CST Studio Suite to Saber (figure 1). The exchange of data is based on the standard Touchstone format for scattering parameters. 3D field simulation results can be exported from CST Studio Suite with only a few clicks. The export function automatically generates all files needed for system simulation in Saber. Figure 1: Interface workflow. Automotive use case Electronic control devices deployed in the automotive domain typically consist of a chassis, a printed circuit board, and a wiring harness. The use case in this paper will focus on a motor cooling fan developed by Bosch. This design includes a "lead frame" that is used as a conductor rail within the ECU power stack to manage the high currents that cannot be supported by a printed circuit board. In order to create a meaningful and precise model for EMC simulation, it's important to account for the 3D geometric description of the lead frame. The lead frame consists of conductor rails that are spatially arranged within the control unit. The arrangement of the lead frame makes an analytical determination of the coupling of the electromagnetic fields impossible. By means of field simulation in the CST Studio Suite, the coupled electromagnetic fields can be calculated numerically. First, the 3D lead frame structure (figure 2) is imported into the CST Studio Suite using CAD data. CST Studio Suite supports the STEP format as well as other popular CAD formats. CST Studio Suite also supports importing printed circuit board layouts EE Times-Asia eetasia.com Copyright 2013 emedia Asia Ltd. Page 2 of 5

3 directly from established layout tools. After the geometry is defined, ports are assigned at locations where electronic components like MOSFETs or recovery diodes are connected to the lead frame. These ports determine the electric interfaces of the lead frame that are used during system simulation in Saber (figure 3). Figure 2: CAD model of an ECU lead frame in CST Studio Suite. Figure 3: Reconditioning of the lead frame model for Saber model export. The model can be simulated once the geometry and ports are defined. The simulation generates EM field data and s- parameters. Figure 3 illustrates the lead frame model in CST Studio Suite as well as the ports required for scattering parameter calculation. In this case, the ECU chassis is used as the reference link for the ports. After the simulation, CST Studio Suite automatically exports the scattering parameters and other supporting files to Saber. The process can be repeated for other components required for ECU system simulation. The complete ECU system model is then created in Saber in accordance with the real-world measurement setup. It consists of a battery, a network model, wiring, and the actual control device. High-quality simulation requires highfidelity models, including parasitics of the components mounted on the circuit board and of the control device. An accurate capacitor model, for example, requires the capacitance as well as parasitic resistance and inductance. The same modelling precision applies to all other components like inductors, diodes or MOSFETs. Board level conductors are also described by parasitic resistors and inductors. Figure 4 shows a partial ECU system model for EMC simulation in Saber. Special attention should be paid to the simulation and analysis of the free-wheeling circuit. The system model includes the test setup, MOSFET driver, engine, and power unit of the control device. The lead frame model has been ported to the system model as a s-parameter black box, generated by the field solver, and is connected to the power semiconductor models that reside on the circuit board. Simulation highlights noise spectrum Fast AC simulation and analysis is performed to calculate the conducted emission. The model is stimulated between the drain and the source pins of the MOSFET. It is necessary to simulate frequencies between 100kHz and 200MHz to determine the effects of AM and FM interference, 100kHz-30MHz and 30MHz-200MHz respectively. The plot shown in figure 5 depicts one of the system simulation results. The purpose of the simulation is to analyse the resonance frequencies and test system performance against voltage amplitudes that are specified by the automobile manufacturers. Amplitudes and resonance frequencies of the control device can be read directly from figure 5. The value added by the simulation partitioning methodology emerges upon comparison of results from two different EMC modelling approaches. The blue line in figure 5 represents the system simulation results for the lead frame exclusively modelled using parasitic resistances and inductors. The red line represents the system simulation in Saber using a S-parameter model of the lead frame that has been generated and exported from CST Studio Suite. When using the scattering parameter model, the emission initially decays more significantly in the FM range and then gets worse at higher frequencies. In contrast, the simplified model with parasitic resistors and inductors EE Times-Asia eetasia.com Copyright 2013 emedia Asia Ltd. Page 3 of 5

4 predicts a steady decrease beyond 100MHz. The differences between both approaches are due to the high frequency EMC effects caused by the 3D geometry of the lead frame. The S-parameter simulation results of the lead frame are verified by measured data. Figure 4: Discrete EMC system model for simulation in Saber. Figure 5: Comparison of conducted emission within the vehicle electrical system. This example demonstrates the impact of a 3D structure on noise emissions generated by the control unit as well as the need for accurate component models within an EMC system simulation. The interface between Saber and CST Studio Suite provides designers with an important bridge between tools that can be used to generate high-quality simulation results for reliable simulation-based design decisions. Synopsys and CST plan to enhance the interface and adapt it to further designer' needs. This article originally appeared in Elektronik.Copyright Elektronik automotive 12/2012, WEKA FACHMEDIEN GmbH. EE Times-Asia eetasia.com Copyright 2013 emedia Asia Ltd. Page 4 of 5

5 References [1] EU framework directive 2007/46/EC, [2] ECE-R10 (Revision 3), United Nations Economic Commission for Europe, [3] "Simulation Tool Chain for the Estimation of EMC Characteristics of ECU Modules," Carsten Schnherr (Robert Bosch GmbH), Synopsys User Group 2008 About the authors M. Eng. Stefan Heimburger studied electrical engineering at the University of Offenburg, Germany. After graduating, he spent five years working in the department of advanced technology development at Bosch in B hler Çtal. After that he joined the product division for actuator systems at Bosch for actuator system development. Since 2007 he has been responsible for network simulation within the EMC group. Dr. rer. nat. Andreas Barchanski studied physics at the Technical University of Darmstadt, Germany, and obtained his Ph.D. in the area of numerical field calculation in After that he joined CST AG as an application engineer. Since 2012 he has been the market development manager for the EMC market. MBA & M. Eng. Thorsten Gerke studied Mechatronics Engineering at the University of Duisburg, Germany. In 2011 he obtained his MBA in General Management from the FOM Business School in Munich. He worked at Bosch in Germany in the field of automotive cranking systems. At Synopsys he is a senior manager for Marketing and Business Development. He has more than 12 years of experience in automotive electronics and has authored over 40 international publications. EE Times-Asia eetasia.com Copyright 2013 emedia Asia Ltd. Page 5 of 5